Leukemia and Lymphoma Metastases - Society for Neuro-Oncology - a

362

15

Leukemia and Lymphoma Metastases

LISA M. DEANGELIS

Central nervous system (CNS) metastases can occurwith any primary systemic cancer, but some primarycancers such as melanoma have a specific predilectionfor the CNS. Brain metastasis is the most common CNSmetastasis, occurring in 15% of all cancer patients(Posner, 1995). Leptomeningeal metastasis is lesscommon, 3% to 8%, and epidural metastasis occurs inapproximately 5% of cases (Posner, 1995; Byrne andWaxman, 1990). Leukemias and lymphomas do metas-tasize to the nervous system but rarely involve brain pa-renchyma and more characteristically involve the lep-tomeninges. Although epidural metastases do notrepresent nervous system metastases because they oc-cur outside of the CNS, they typically have a neurologicpresentation and for that reason are considered here.

The overwhelming majority of CNS metastases aredue to solid tumors rather than to lymphoreticularmalignancies. Lymphoma accounts for only 10% ofepidural metastases whereas solid tumors account forthe remaining 90% (Posner, 1995; Byrne and Wax-man, 1990); leukemia rarely causes epidural disease(Bower et al., 1997; Kataoka et al., 1995). In con-tradistinction, the lymphoreticular malignancies ac-count for a preponderance of patients with lep-tomeningeal metastases. The overall incidence isdifficult to ascertain because leukemias and lym-phomas are often excluded from most series, but ap-proximately 24% of patients with leptomeningeal me-tastasis have non-Hodgkin’s lymphoma (NHL) (Olsonet al., 1974). Therefore, the pattern of CNS metas-tases from lymphoma and leukemia is different fromthat of solid tumors, and the differential diagnosis ofthese entities is different for patients with lym-

phoreticular malignancies. For example, leptom-eningeal metastasis can mimic vincristine peripheralneuropathy, which is common among patients withlymphoma or leukemia. Patients with lymphoreticu-lar malignancies are particularly vulnerable to op-portunistic infections, which can mimic metastasis.Finally, isolated CNS metastasis is far more commonwith lymphoma or leukemia than with solid tumorswhere CNS disease typically occurs in the setting ofwidespread systemic metastases.

Systemic therapy of leukemia and lymphoma canbe highly effective and can eradicate extra-CNS dis-ease. However, microscopic tumor within the CNSmay be protected from circulating systemic chemo-therapy by the blood–brain barrier. This disease canprogress while the patient is in remission systemi-cally, leading to an isolated CNS relapse. This patternof recurrence is characteristic of the leukemias andlymphomas, making them different from the solid tu-mors and warranting special consideration.

EPIDURAL METASTASES

Epidural metastases are seen in 3% to 5% of patientswith systemic NHL (Levitt et al., 1980; Mackintosh etal., 1982; Raz et al., 1984). Epidural lymphoma canbe the presenting manifestation of disseminated NHL,or can be an isolated site of disease, which accountsfor approximately 1% of patients with NHL (Lyons etal., 1992; Gilbert et al., 1978). Epidural tumor oc-curs primarily in those patients with intermediate- tohigh-grade subtypes and in those with advanced dis-

3601_e15_p362-374 2/19/02 8:56 AM Page 362

ease (i.e., stage III or IV). Occasionally, the devel-opment of a complication such as epidural metasta-sis heralds the transformation of a previously low-grade or indolent neoplasm into a higher grademalignancy or may be the initial manifestation of theillness. The development of epidural metastases tendsto occur in those patients with bone metastases, par-ticularly vertebral metastases, and in those who haveparaspinal nodal involvement. It has also been asso-ciated with retroperitoneal adenopathy and, in someseries, with bone marrow infiltration.

Epidural metastasis is a very rare complication ofany type of leukemia. It can be seen as a consequenceof paraspinal chloroma formation in patients withacute myeloblastic leukemia (AML). It has a presen-tation, diagnosis, and treatment identical to epiduralmetastasis from NHL, and the following discussioncan be applied to these unusual patients.

Clinical Features

The clinical features of epidural metastasis from lym-phoma are not substantially different from those seenin solid tumors and described in Chapter 14. The pre-dominant clinical symptom is back or neck pain(Byrne and Waxman, 1990; Gilbert et al., 1978; Pos-ner, 1987), which is present in 95% of patients withepidural metastasis. Usually the first symptom, it of-ten predates the development of neurologic deficitsby months. The pain is typically thoracic, an unusualsite of pain due to degenerative disease, because 80%of epidural metastases are in the thoracic spine. Mostpatients with epidural metastasis from solid tumorspresent first with back pain, which may develop aradicular component as the disease progresses. Thisoccurs because the metastasis initially originates inthe bone, usually the vertebral body, and then growsoutside of the bone to involve paraspinal structuresand cause nerve root compression.

In contrast, NHL more commonly involves theepidural space by tumor growing from the paraverte-bral area directly through the intervertebral foramen,causing spinal cord compression. For this reason, thereis less back pain from bone destruction. The pain morecommonly has a radicular component or may even bereferred within the dermatomal distribution of the com-pressed root, which can lead to misdiagnosis. Radicu-lar pain down a limb or across the trunk may, in fact,be the first indication of an epidural tumor from NHL.Unlike solid tumors, NHL can occasionally metastasize

directly to the epidural space without bone or para-vertebral involvement. These lesions may be asympto-matic and initially detected on body CT scans done tocompletely stage the patient’s NHL. If suggested on CTscan, a comprehensive evaluation with magnetic reso-nance imaging (MRI) (see below) is essential to es-tablish the diagnosis.

In order of frequency pain is followed by leg weak-ness, which occurs in approximately 50% of patients,and may be accompanied by sensory dysfunction ina comparable proportion. Sphincter dysfunction isseen in about 20% of patients. The back pain ofepidural cord compression is characterized by pro-gressive severity as well as increased severity whenthe patient lies down, in contrast to pain from de-generative spinal disease, which characteristically im-proves upon recumbency. In addition, pain that in-tensifies with cough, sneeze, or Valsalva stronglyindicates compression of the spinal cord, which istransiently intensified with the increase in intraspinalpressure that occurs with these maneuvers. Some-times these features can alert the physician that theback pain is due to something more serious than thecommon, benign causes of back pain.

Diagnosis

The best and only test necessary to establish a diag-nosis of epidural metastasis is a spinal MRI (Jordanet al., 1995). This should be done without intravenouscontrast material (i.e., gadolinium), which can actu-ally obscure the diagnosis and make it more difficultto see tumor. Magnetic resonance imaging can visu-alize the entire spine noninvasively and identifyepidural tumor at any level (Fig. 15–1). It is partic-ularly useful for patients with lymphoma in whom tu-mor can enter the epidural space via the interverte-bral foramen and not involve or destroy bone. Thisis a major limitation of plain films and bone scans,which can only identify sites of bony destruction. Evenif a bone metastasis is present, these techniques donot indicate whether or not the disease has pro-gressed to involve the epidural space. Furthermore,they can be negative in the face of significant bone in-volvement with epidural tumor causing spinal cordcompression. Any patient with NHL who has signifi-cant, progressive back pain should be considered forspinal MRI even in the absence of neurologic deficits.

Another feature of MRI is that it easily images theentire spine. This is essential as multilevel epidural

Leukemia and Lymphoma Metastases 363

3601_e15_p362-374 2/19/02 8:56 AM Page 363

disease occurs in about 5% of patients with anepidural metastasis. Consequently, if MR images areobtained, an epidural metastasis identified, and onlya portion of the spinal column visualized, then thepatient should return to the scanner to completeimaging of the remainder of the spine.

Some patients are unable to undergo MR imag-ing because they have a pacemaker or other devicethat prohibits them from being in a high magneticfield. A computed tomography (CT) myelogramshould be performed for such patients. If a com-plete block is identified with dye introduced into thelumbar space, then a C1–C2 puncture should beperformed to introduce dye from above to definethe upper limit of the epidural tumor. This is par-ticularly important in NHL in which the disease cangrow extensively in the rostral caudal direction onceit has reached the epidural space. Accurate identi-

fication of the full extent of tumor is critical for treat-ment planning.

Initial Management

Many patients with an epidural metastasis are easilyidentified clinically. They have severe progressiveback pain accompanied by neurologic symptoms andsigns suggestive of a myelopathy. For such patients,dexamethasone is often administered even beforeneuroimaging is obtained. Dexamethasone rapidly re-lieves the pain of spinal cord compression and mayfacilitate neurologic recovery. Experimental data andsubstantial but retrospective clinical data suggest adose–response relationship between corticosteroidsand control of back pain associated with epidural tu-mor (Posner, 1995). The pain can be substantiallyameliorated within hours of drug administration,which can facilitate the patient’s ability to tolerate anydiagnostic procedure, especially an MRI scan. Typi-cally, an intravenous bolus of dexamethasone is ad-ministered. Clinical data support the use of a very highinitial dose, 100 mg, to rapidly relieve back pain(Loblaw and Laperriere, 1998). With the exceptionof patients with NHL, for those patients with knowncancer, particularly solid tumors, this is a very rea-sonable approach.

Corticosteroids are a well-recognized, effectivechemotherapeutic agent for the treatment of NHL. Be-cause they can cause rapid cell lysis, tumor can dis-appear very quickly after their administration (Pos-ner et al., 1977). Consequently, it is essential thatneuroimages be obtained before the dispensation ofany corticosteroids to NHL patients suspected of hav-ing epidural tumor. Their pain should be managedwith narcotic analgesics to facilitate performing thenecessary neuroimaging. Once the MRI scan is com-plete and an epidural metastasis has been identified,administration of dexamethasone is appropriate.

This approach is very straightforward for patientswith known NHL, but it becomes more complicatedfor patients whose malignancy presents for the firsttime as an epidural mass. For such patients, MR im-ages are obtained first, and there is usually no con-sideration given to administering corticosteroids be-fore identification of an epidural mass. Once such amass is seen on an MRI scan, however, corticos-teroids are usually given immediately. If the mass isa lymphoma, one can see rapid resolution of the le-sion. If tissue has not yet been obtained for diagno-

364 CANCER METASTATIC TO THE CENTRAL NERVOUS SYSTEM

Figure 15–1. Magnetic resonance image of the spinedemonstrating ventral epidural lymphoma extending from T2to T5. Note the preserved vertebral bodies and absence of bonedestruction.

3601_e15_p362-374 2/19/02 8:56 AM Page 364

sis, the opportunity to confirm the diagnosis patho-logically is thus lost, and appropriate treatment is de-ferred, resulting in a significant delay for the patient.Despite their clinical response to the corticosteroids,patients must be tapered off the drug to allow the dis-ease to declare itself once again so that tissue can beobtained for biopsy. Not only does this delay defini-tive treatment, but also puts the patient at substantialrisk of progressive neurologic compromise from re-current epidural metastasis. It is essential that theseissues be considered before “standard therapy” is ad-ministered on a routine basis.

Treatment

Once the diagnosis is established, treatment of epiduralmetastasis should be implemented as rapidly as possi-ble. Treatment may involve any one of the three majoranticancer therapeutic modalities: radiotherapy (RT),surgery, and chemotherapy. The choice of treatment,or combination of therapies, depends on the patient’sclinical and neurologic condition, his or her priortreatment for the underlying lymphoma, and any priortherapy for epidural metastasis. Rapid institution oftreatment is imperative as a patient’s neurologic func-tion can deteriorate precipitously. A general rule ofthumb for most patients with spinal cord compressionis that if they are ambulatory at diagnosis, they remainambulatory after treatment, but if they are nonambu-latory at diagnosis, they rarely regain the ability to am-bulate independently (Posner, 1995).

Radiotherapy

Radiotherapy is the most common and effective mo-dality for the treatment of epidural spinal cord com-pression (Maranzano et al., 1991; Bilsky et al., 1999).It is easily administered and highly effective, particu-larly for a radiosensitive primary tumor such as lym-phoma. A complete spinal MRI will define the rostralcaudal extent of the epidural metastasis. Typically, weadminister radiotherapy to a port encompassing thearea of tumor plus two vertebral bodies superior andinferior to the tumor margin. The usual course oftreatment is 300 cGy for 10 fractions, for a total of30 Gy. Patients should receive corticosteroids beforeand during RT to minimize exacerbation of neuro-logic problems from edema engendered by the treat-ment, but for select patients steroids are not requiredduring RT (Maranzano et al., 1996).

Radiotherapy is particularly effective for NHL fortwo reasons: (1) Lymphoma is a highly radiosensi-tive neoplasm so that focal RT can be very effectivein relieving a spinal cord compression from epiduralmetastasis; and (2) because lymphoma frequently involves the epidural space by growing through theintervertebral foramen, or metastasizing directly tothe epidural compartment, bone destruction is a lessprominent feature of epidural metastases from NHL.Epidural spinal cord compression is typically causedby the tumor itself and is a consequence of soft tis-sue compression rather than bone compression sothat RT is more likely to relieve spinal cord com-pression in this circumstance.

Side effects from RT can include myelosuppres-sion, particularly if the patient is heavily pretreatedwith chemotherapy or a long expanse of spine mustbe included in the port of RT. Patients can also de-velop gastrointestinal irritation from RT to the lowerspine or mucositis from cervical RT.

Surgery

Surgery is rarely the first line of treatment for patientswith spinal cord compression from lymphoma (Byrneand Waxman, 1990). It is used initially when a tissuediagnosis has not been made. Surgery can establishthe diagnosis and also decompress the spinal cord.Even if gross total excision of the disease seems tohave been accomplished in such patients, postoper-ative RT is appropriate to avoid local recurrence andsubsequent recompression of the spinal cord.

Surgery may be appropriate for patients who areexperiencing spinal cord compression in a previouslyirradiated location. For those patients who are notcandidates for a second course of RT (Schiff etal.,1995), surgery may improve or at least maintainneurologic function (Bilsky et al.,1999; Sioutos et al.,1995; Klekamp and Samii, 1998). The surgical ap-proach depends on the location of the tumor mass.If the tumor has arisen from a vertebral body metas-tasis and is compressing the cord anteriorly, a verte-brectomy from an anterior approach may be most ap-propriate. Data suggest that such patients who havesevere neurologic impairment may regain sphinctercontrol and leg strength when a complete decom-pression is achieved by anterior resection. However,disease that has arisen in the paravertebral locationor more posteriorly may be more amenable tolaminectomy, which would allow for tumor removal


3601_e15_p362-374 2/19/02 8:56 AM Page 365

and direct decompression of the spinal cord.Laminectomy for patients with disease located ante-riorly probably does not improve neurologic outcomeand does not effectively treat the tumor.

Surgery should be reserved for patients in goodpreoperative condition who have systemic disease thatis controlled or controllable and who do not havemultilevel epidural tumor (Sioutos et al., 1995;Klekamp and Samii, 1998). Surgical complicationsinclude worsening neurologic deficit, wound dehis-cence or infection (particularly in those who requiresustained doses of corticosteroids), and delayedhardware disruption, which often heralds tumor re-growth.

Chemotherapy

Chemotherapy is rarely the first line of treatment forepidural metastasis. However, it has been shown tobe effective for those patients whose epidural tumorswere identified during their extent of disease evalua-tion at initial presentation (Lyons et al., 1992; Oviattet al., 1982; Wong et al., 1996). These patients typi-cally have an epidural site of disease identified on theinitial body CT scan done to evaluate intrathoracicand intra-abdominal disease. Epidural tumor is thenconfirmed by spinal MRI; however, patients may beasymptomatic or have minimal neurologic sympto-matology. These patients usually require combinationchemotherapy as an initial treatment for systemic lym-phoma. For such patients, chemotherapy can be ad-ministered and the epidural disease monitoredclosely. Typically, the epidural tumor responds in thesame fashion as the rest of the systemic disease. Forthose patients whose epidural disease does not re-spond, focal RT can be administered.

Although epidural tumor presents with neurologicsymptoms and signs referable to the spinal cord, it isimportant to remember that epidural disease existsoutside of the CNS and is not behind the blood–brainbarrier. Systemically administered chemotherapy is aseffective against disease in this location as in any sys-temic location. The choice of drugs should be basedon the optimal regimen likely to be effective againstthe systemic lymphoma.

Chemotherapy can also be used for patients whosedisease has developed in a previously irradiated siteor for whom surgery is not an option or has alreadyfailed. However, for such patients who have heavilypretreated disease and for whom prior chemother-

apy has often been administered, the probability ofan excellent response from such treatment is sub-stantially less than at initial therapy.

LEPTOMENINGEAL METASTASES

Leptomeningeal metastases develop as a complicationin 4% to 11% of patients with systemic NHL and inapproximately 10% of patients with leukemia (Table15–1) (Posner, 1995; Olson et al., 1974). In patientswith NHL, the incidence is higher among those withhigh-grade and widespread disease. In leukemia pa-tients, the incidence varies widely with type of leuke-mia, reaching a peak incidence of 56% at autopsy inthose with acute lymphocytic leukemia (ALL). Thiswas particularly true before the availability of pro-phylactic intrathecal chemotherapy (Price and John-son, 1973). The development of vigorous systemicand CNS therapies has markedly decreased the inci-dence of meningeal leukemia in both ALL and acutemyelocytic leukemia (AML) (Barcos et al., 1987).Currently, the incidence of CNS relapse is 2.2% inAML and 4.3% in ALL (Castagnola et al., 1997; Starket al., 2000).

Clinical Features

The hallmark of leptomeningeal metastasis is multi-focal involvement of the CNS (Wasserstrom et al.,1982; Balm and Hammack, 1996). The disease pri-marily involves three main regions of the CNS: cra-nial nerves, the cerebrum, and spinal compartment(Table 15–2). Patients may present with symptomsand signs involving one or all of these locations and


Table 15–1. Frequency of Leptomeningeal Metastasis

No. of No. ofAutopsies Metastases (%)

Leukemia 287 28 (10)

ALL 87 21 (24)

AML 104 5 (5)

Lymphoma 309 15 (4)

Hodgkin’s 119 2 (2)

Non-Hodgkin’s 190 13 (7)

ALL, acute lymphocytic leukemia; AML, acute myelogenous leukemia.

Source: Adapted from Posner (1995).

3601_e15_p362-374 2/19/02 8:56 AM Page 366

generally have more neurologic signs on examinationthan symptoms. This discrepancy is often the first cluethat meningeal tumor is present.

Common symptoms are facial weakness, facialnumbness and diplopia (Posner, 1995; Levitt et al.,1980). Radicular symptoms are most commonly ob-served in the legs, and numbness or weakness maybe bilateral but is often asymmetric. Bowel and blad-der disturbances are frequent. Cerebral symptoms areusually due to raised intracranial pressure and com-municating hydrocephalus caused by tumor impair-ing absorption of cerebrospinal fluid (CSF) over thecerebral convexities. Headache and mental statuschanges are the most common cerebral symptoms.Seizures and ataxia are infrequent, occurring in fewerthan 10% of patients. Lateralizing symptoms andsigns, including hemiparesis, aphasia, or a visual fielddeficit, are not seen with leptomeningeal metastasisunless there is an accompanying parenchymal lesion,such as a brain metastasis; or leptomeningeal tumorhas caused vascular occlusion leading to a stroke.

Pain is a variable accompaniment to leptomenin-geal tumor. The cranial neuropathies are typicallypainless, except facial pain is reported occasionally.The radicular symptoms and signs may be painless,which is often a clue that the cause is leptomeningealtumor rather than epidural tumor, which is almostalways painful. However, radicular pain can be aprominent symptom of leptomeningeal metastasis andcan be difficult to treat. When there is significantradicular, back, or neck pain, epidural tumor is themost important differential diagnostic consideration.

Diagnosis

The diagnosis of leptomeningeal lymphoma or leu-kemia has usually required the demonstration of tu-

mor cells in the CSF, which are almost always ab-normal in the presence of leptomeningeal metastasis.Positive cytology is, however, observed in only 50%of patients with documented leptomeningeal metas-tasis from solid tumors on the first lumbar puncture(Wasserstrom et al., 1982). Repeated spinal taps arefrequently needed to demonstrate the presence of tu-mor cells, and positive cytology results can be ob-tained in 90% of patients with three lumbar punc-tures. However, with NHL leptomeningeal metastasis,the CSF yields a positive cytologic examination in 88%of patients with two lumbar punctures (Recht et al.,1988).

Routine studies of CSF are less helpful in patientswith leukemia and NHL than in those with solid tu-mors (Posner, 1995; Wasserstrom et al., 1982; Rechtet al., 1988). Cerebrospinal fluid protein concentra-tion is elevated in approximately 60%, but rarelyabove 200 mg/dL. The CSF glucose level may be de-pressed, but only in a minority of patients. The CSFcell count is usually elevated and may be composedof tumor cells and reactive lymphocytes, making a cy-tologic distinction between the two very difficult insome patients. These abnormalities are seen in manypatients with leptomeningeal lymphoma and leuke-mia, as they are in patients who have this complica-tion from solid tumors. The exception is lep-tomeningeal leukemia, which can be present inotherwise completely normal CSF (Tubergen et al.,1994; Mahmoud et al., 1993). In particular, the cellcount may be normal if patients are pancytopenicfrom either their disease or its treatment. Therefore,vigilance in the cytologic examination is essential forpatients suspected of this process. Sending large vol-umes of CSF to the cytopathologist with rapid fixationof the specimen can increase the yield. In addition,sampling CSF from a location close to the area of clin-ical symptoms also improves yield. For example, pa-tients with lumbar radicular symptoms have the high-est incidence of positive cytology when CSF samplesfrom a lumbar puncture are used (Rogers et al.,1992). However, those with cerebral symptoms orcranial neuropathies have a higher yield when CSF isobtained from a C1–C2 puncture or a ventricularsample when a ventricular reservoir is already inplace.

Cerebrospinal fluid tumor markers can occasion-ally be helpful in identifying tumors (Schold et al.,1980; Oschmann et al., 1994; DeAngelis, 1998), butbecause there are no specific tumor markers for leu-


Table 15–2. Symptoms and Signs of LeptomeningealMetastases in Lymphoma

Radiculopathy 44%

Impaired mental function 42%

Cranial neuropathy 36%

Headaches 23%

Seizures 3%

None (positive CSF cytology only) 11%

Source: Adapted from Recht et al. (1998).

3601_e15_p362-374 2/19/02 8:56 AM Page 367

kemia and lymphoma, they are more effective foridentifying solid tumors. �2 Microglobulin is often el-evated in CSF in lymphoma and occasionally in leu-kemia, but is nonspecific and can be elevated in anyinflammatory condition associated with a CSF pleo-cytosis. While this is also true of the nonspecific mark-ers �-glucuronidase and LDH isoenzymes, thesemarkers can nevertheless be useful in some patients(Lossos et al., 2000). Vascular endothelial growthfactor has recently been shown to be predictive in pa-tients with leptomeningeal metastasis from solid tu-mors; it may also prove valuable in hematologic malignancies (Stockhammer et al., 2000). Flow cy-tometry and molecular markers are helpful if ade-quate cells are available and the molecular pheno-type is known (van Oostenbrugge et al., 1998; Cibaset al., 1987; Rhodes et al., 1996).

Demonstration of tumor cells in the CSF is not theonly way to establish a diagnosis of leptomeningealmetastasis. Gadolinium-enhanced MRI of the neuraxissometimes reveals findings that are so characteristicof leptomeningeal tumor as to be diagnostic (Rode-sch et al., 1990; Freilich et al., 1995). Prominent en-hancement and enlargement of cranial nerves due totumor infiltration, nodules adherent to the caudaequina, large subarachnoid masses compressing thespinal cord, and prominent enhancement coating thesurface of the brain extending deep into sulci are alldefinitive neuroradiologic features of tumor in the

subarachnoid space in patients known to have can-cer (Fig. 15–2). The presence of such findings, evenin the absence of a positive CSF cytologic examina-tion, can establish the diagnosis and be sufficient toinitiate treatment.

These findings are not manifest in every patientwho has leptomeningeal tumor. Normal neuroradio-logic studies do not exclude leptomeningeal tumor,which is particularly problematic for patients withleukemia and lymphoma who have a lower incidenceof neuroradiologic abnormalities than those withsolid tumors. Furthermore, cranial imaging that re-veals a pattern of miliary brain metastases with smalllesions evident in the sulci of the brain, or superfi-cially on the cortex, may suggest the presence of lep-tomeningeal tumor. These findings are exceedinglyrare in lymphoma and leukemia. All patients sus-pected of having leptomeningeal tumor, and those inwhom tumor has been confirmed on CSF cytologicexamination, should undergo complete imaging of theneuraxis with gadolinium to delineate areas of focalor bulky disease, which may require focal RT as partof the treatment plan.

While the diagnosis of leptomeningeal tumor canbe extremely difficult to make in any circumstance,the situation is particularly challenging for patientswith leukemia and lymphoma. The tendency of thesetumors to grow in sheets and not to form nodulesmakes diagnosis difficult because the incidence of


Figure 15–2. Gadolinium-enhanced MRI demonstrating bilateral enhancement and infiltration of the trigeminal nerves by lym-phoma.

3601_e15_p362-374 2/19/02 8:56 AM Page 368

bulky disease, and, therefore, detectable disease onneuroimaging, is much lower. Furthermore, the inci-dence of leptomeningeal metastases is higher in thesetumor types than in any other, making the need forrecognition and aggressive treatment a common con-cern. Most importantly, vigorous treatment of lep-tomeningeal metastases in patients with leukemia andlymphoma can lead to prolonged remission and,sometimes, even cure. Consequently, it is imperativeto diagnose these tumor types early.

When the diagnosis is not established on CSF anal-ysis or neuroimaging, the clinician may deduce it byprocess of elimination. Imaging helps to exclude al-ternatives such as epidural or vertebral bone metas-tases, brachial or lumbosacral plexopathy, andparenchymal brain pathology. Laboratory work canexclude metabolic causes of lethargy or seizure. Themedical history can usually indicate whether cranialneuropathies can be attributed to drugs such as vin-cristine. When all alternative diagnoses have been ex-cluded and the patient has a characteristic presenta-tion such as cranial neuropathy, some experiencedclinicians will treat leukemia or lymphoma patientsfor leptomeningeal metastasis even in the absence ofdiagnostic confirmation.

Initial Management

Neurologic metastatic complications frequently lead thephysician to initiate corticosteroid treatment immedi-ately. Leptomeningeal metastasis from solid tumorsrarely responds to corticosteroids unless the patienthas markedly increased intracranial pressure. How-ever, in lymphoma and to a lesser extent leukemia, cor-ticosteroids can provide symptomatic relief, particu-larly from pain. This is because corticosteroids canfunction as a chemotherapeutic agent in lymphoma andcause tumor lysis, an important issue when the diag-nosis is suspected but not yet confirmed. Prematureadministration of corticosteroids can give false-nega-tive CSF cytologic examination and neuroimaging re-sults. Corticosteroids should be reserved until the di-agnosis has been established, at which time they mayprovide symptomatic relief. In the absence of clinicalimprovement, steroids should be rapidly tapered andthen discontinued. Unlike brain or epidural metastases,corticosteroids are not required for irradiation of leptomeningeal tumor because there is no focal in-volvement or compression of the nervous system, in-citing significant local edema or mass effect.

Treatment

Therapy should begin immediately after confirmationof the diagnosis of leptomeningeal metastasis. Thegoals of treatment are not only to prolong life but alsoto minimize neurologic disability. Rapid institution oftherapy can halt the progression of neurologic dys-function and, if the disease has not been long–stand-ing, can often reverse some neurologic disability.Therapeutic choices include RT, systemic chemo-therapy, and intrathecal chemotherapy. Which treat-ment is selected depends on the location and extentof leptomeningeal involvement as well as the patient’ssymptoms.

Radiotherapy

Radiotherapy can be a highly effective treatment, fre-quently causing rapid relief of pain and occasionallyreversal of neurologic symptoms (Hanssens et al.,1998). It is usually delivered to focal areas of bulkydisease seen on MRI and to symptomatic areas. Forexample, patients with lumbosacral radiculopathywould receive RT to the cauda equina, whereas thosewith cranial neuropathies would receive either wholebrain or skull base RT. Radiotherapy is usually de-livered in 300 cGy fractions for a total of 3000 cGy.Often, its effect can be substantial and durable, but itis not curative (Mackintosh et al., 1982; Hanssens etal., 1998; Gray and Wallner, 1990).

The major limitation of RT is that it is adminis-tered focally, leaving large areas of the subarachnoidspace untreated. Because the CSF circulates along theneuraxis, tumor cells can be carried by bulk flowfrom one region to the other. Tumor cells can thusfloat in and out of the port of RT, never receiving asufficient dose. Also, large areas of the neuraxis areuntreated by focal RT. Neuraxis RT can treat the en-tire CSF compartment, but craniospinal RT is quitemorbid, resulting in esophagitis and enteritis in manypatients. In addition, treatment of the entire spinalaxis often results in significant myelosuppression,particularly in heavily pretreated patients who previ-ously received substantial chemotherapy. This se-quela often causes interruption of treatment and, ifsevere, can necessitate transfusion or result in neu-tropenic infection or thrombocytopenic bleeding.Even focal spinal RT can occasionally result in de-pressed blood counts in some patients, although thecondition is usually easily managed.


3601_e15_p362-374 2/19/02 8:56 AM Page 369

Intrathecal Chemotherapy

Intrathecal chemotherapy delivers drug into the sub-arachnoid space to treat the entire CSF compartment.Most systemically administered chemotherapeuticagents do not achieve sufficient concentration in theCSF to treat tumor cells, so drug must be instilled di-rectly into the CSF. This is a safer method of treatingthe entire CSF than neuraxis RT. However, the num-ber of drugs that can be safely administered directlyinto the CSF is limited, and the most commonly usedagents are methotrexate, cytarabine, and thiotepa.These agents have a relatively narrow antitumor spec-trum but can be effective in treating both lymphomaand leukemia. Other agents such as etoposide havebeen used experimentally with some efficacy but havenot been adopted for routine use (van der Gaast etal., 1992; Champagne and Silver, 1992; Berg et al.,1992).

Intrathecal chemotherapy can be administered ei-ther by repeated lumbar punctures or by placementof a ventricular catheter with an Ommaya reservoir(Berweiler et al., 1998), which allows easy accessi-bility to the subarachnoid compartment and resultsin better disease control (Shapiro et al., 1975; Bleyerand Poplack, 1979). Use of a reservoir has three ma-jor advantages over repeated lumbar punctures. Drugdelivered into a reservoir has better distributionthroughout the CSF than drug introduced into thelumbar space (Shapiro et al., 1975). Even when alumbar puncture is successful and the CSF is reached,injection of drug via a spinal needle results in instil-lation of the drug into the epidural space in approx-imately 10% of patients (Larson et al., 1971). Finally,the reservoir is much easier on the patient, and drugadministration is less time consuming for staff.

However, the reservoirs can pose occasional diffi-culties and complications. They have a low incidenceof infection, but, when infected, may require removalto clear infection, which is most commonly due toskin organisms such as coagulase-negative Staphylo-coccus or Proprionibacterium species. The reser-voirs may become obstructed. If this develops, thereservoir should not be used as the drug may leakout of the reservoir catheter and into the surround-ing brain, causing an area of focal encephalomalaciaor a sterile abscess that can result in focal neurologicdeficits. Patients with raised intracranial pressure areparticularly vulnerable to this complication. In addi-tion, patients with hydrocephalus or any impairmentof CSF flow should not have a reservoir placed.

Drug is distributed along with the bulk flow of CSF.If there is obstruction to CSF flow, the drug will betrapped in one area of the neuraxis, leaving other re-gions untreated and causing neurotoxicity where theconcentration is high for prolonged periods of time(Glantz et al., 1995; Mason et al., 1998). 111Indiumflow studies can ascertain with a high degree of ac-curacy whether the CSF flow is normal or not (Cham-berlain, 1998). The 111Indium should be adminis-tered by the same route as the drug, either via anOmmaya reservoir or by lumbar puncture. The pa-tient is scanned for distribution of the 111Indiumthroughout the neuraxis and for reabsorption overthe cerebral convexities. Areas of bulky disease seenon MRI, such as large subarachnoid nodules in thespine, usually impair CSF flow at that level. The flowcan occasionally be restored after focal RT has beenadministered to the area. There is controversy re-garding obstructions, so-called physiologic obstruc-tion, seen on 111Indium studies where neuroimagingis negative (Glantz et al., 1995; Mason et al., 1998;Chamberlain, 1998). Some authors recommend ra-diating areas of reduced CSF flow; restoration of flowoccasionally occurs. Others, however, remain un-convinced that this is a significant phenomenon andare concerned that RT can cause toxicity.

For drugs administered into the CSF, the doses arefixed and should not be calculated on a meter squarebasis. The volume of CSF is the same in all individu-als over the age of 4 years, and it does not fluctuatewith body size (Pfefferbaum et al., 1994). Doses ofthe commonly used agents are indicated in Table15–3. When delivered into a reservoir, drug shouldbe infused slowly because rapid administration canproduce raised intracranial pressure and hypoten-sion. If there is difficulty in removing CSF from thereservoir, placing the patient in the Trendelenburgposition may facilitate CSF withdrawal.

When intrathecal methotrexate is used, oral leu-covorin should be given for the following 4 days at adose of 10 mg po b.i.d. This is to protect the gas-trointestinal tract and the bone marrow from thechronic low-dose systemic exposure that results from


Table 15–3. Doses of Intrathecal Chemotherapy

Methotrexate 12 mg

Cytarabine 40–60 mg

Thiotepa 10 mg

3601_e15_p362-374 2/19/02 8:56 AM Page 370

reabsorption of drug from the subarachnoid spaceinto the bloodstream.

The optimal duration and schedule of intrathecaltreatments for leptomeningeal metastasis is unknown.The schedule has been derived largely from empiricinformation and from what is known about the phar-macokinetics of drug in the CSF (Mackintosh et al.,1982). Methotrexate is the best-studied agent. Itrapidly achieves high concentrations after an in-trathecal dose and maintains therapeutic concentra-tions in the CSF for at least 24 hours (Shapiro et al.,1975), which has led to initial treatment on a twice-a-week schedule so that the tumor receives thera-peutic concentrations for a substantial period of time.If the patient appears to be responding to intrathecaldrug, as assessed by improvement in the CSF and theabsence of clinical deterioration, then after severalweeks of giving the drug twice a week the frequencyis reduced to once a week for an additional 3 to 4weeks. This is followed by a further reduction to everyother week and then a few months of monthly main-tenance therapy, which is then discontinued.

An alternative approach is to use the concentra-tion � time approach where small doses are givendaily for 3 to 5 days (Moser et al., 1999), which pro-duces sustained therapeutic concentrations in the CSFwith a reduced total dose, possibly diminishing therisk of neurotoxicity. This is a superior approach butcan be difficult to administer because of the require-ment for daily injections. A new preparation of cy-tarabine addresses some of these issues. This liposo-mal preparation can be administered intrathecallyonce every 2 weeks (Glantz et al., 1999). This methodreleases drug slowly and can produce therapeuticconcentrations of cytarabine in the CSF for more than1 week in most patients. The consequent need for lessfrequent administration is an improvement in the pa-tient’s quality of life. This preparation may or may notrepresent a substantial therapeutic advantage overstandard cytarabine or methotrexate but that remainsto be ascertained.

Systemic Chemotherapy

Most systemic agents do not penetrate sufficiently intothe subarachnoid space to treat leptomeningealmetastases. However, agents such as methotrexateand cytarabine when given in high doses can achievetherapeutic concentrations in the CSF (Glantz et al.,1998). This is particularly true when tumor involvesthe leptomeninges, which enhances drug penetration.

The advantage of delivering chemotherapy systemi-cally is that sufficient drug concentration can beachieved throughout the subarachnoid space becausethe drug does not have to circulate with the CSF toreach all areas of the subarachnoid compartment. Inaddition, systemically administered drug can reachand penetrate into nodules of tumor in the sub-arachnoid space and into neural structures infiltratedby tumor. Intrathecally administered drug can pene-trate only 5 mm into the tumor and therefore cannotreach these areas of disease.

Other important agents in the treatment of lym-phoma and leukemia such as anthracyclines, vin-cristine, and cyclophosphamide do not achieve suffi-cient penetration into the CSF to effectively treatleptomeningeal metastasis. Consequently, the optionsavailable for systemic chemotherapy are virtually thesame drugs that can be administered directly into theCSF. Nevertheless, data suggest that patients do bet-ter when systemic chemotherapy is included in thetherapeutic regimen of leptomeningeal metastasis(Siegal et al., 1994; Siegal, 1998).

OUTCOME

The prognosis for most patients with leptomeningealmetastasis is poor, with a median survival of 6 to 8months for patients with NHL and 10 months for thosewith AML (Posner, 1995; Castagnola et al., 1997;Recht et al., 1988). It is difficult to control tumor inthe CNS, and also, because leptomeningeal metasta-sis is usually a late complication of either lymphomaor leukemia, the systemic disease is often aggressiveand refractory to treatment by the time metastasis isdiagnosed, and death is related to progressive sys-temic tumor in most patients (Recht et al., 1988).Nevertheless, some patients do respond well to treat-ment and have prolonged survival (Siegal et al.,1994). Furthermore, for patients with isolated CNSrelapse, control or remission of CNS disease can re-sult in a sustained second remission so that vigoroustreatment is warranted. Aggressive therapy can pre-vent neurologic dysfunction even if survival is not pro-longed, which provides a substantial contribution toa patient’s quality of life. Patients who do survive for1 year or longer are, however, vulnerable to late neu-rotoxic effects of treatment (Siegal et al., 1994). Pri-marily restricted to the brain, this is a particular prob-lem for patients who have received whole-brain RTin addition to chemotherapy.


3601_e15_p362-374 2/19/02 8:56 AM Page 371

NEUROTOXICITY

Combining cranial irradiation with systemic and/orintrathecal chemotherapy amplifies the potential ofeach modality to cause neurologic dysfunction. Therisk rises with increasing doses of RT, a rising cu-mulative dose of systemic and/or intrathecal chemo-therapy, and older age. The primary manifestation ofneurotoxicity is memory impairment, which canprogress to a severe dementia in adults or be a sta-tic learning deficit in children. Radiographically, a dif-fuse leukoencephalopathy is seen on MRI with in-creased signal throughout the periventricular whitematter on T2 or FLAIR images. Atrophy and ventricu-lar dilatation are also common features. Occasion-ally, some patients may have amelioration of theirsymptoms with placement of a ventriculoperitonealshunt, but improvement is usually incomplete and canbe temporary. Once neurotoxicity develops, it is a per-manent and irreversible condition. Considerable ef-fort has been devoted to the development of effica-cious, less toxic regimens for CNS prophylaxis,particularly for childhood ALL. Data show that vigor-ous systemic chemotherapy combined with extendedtriple intrathecal chemotherapy can produce controlof CNS disease comparable to cranial RT plus drug(Stark et al., 2000). This type of regimen carries lessrisk of subsequent cognitive impairment.

PRIMARY CENTRAL NERVOUS SYSTEM LYMPHOMA

Clinical Features

Primary CNS lymphomas (PCNSL) represent 3.5% ofprimary brain tumors (Davis and Preston-Martin,1999) and are usually large cell or immunoblasticlymphomas. There is a higher incidence of PCNSL intransplant recipients and patients with acquired im-munodeficiency syndrome (AIDS), but these tumorsare stimulated by latent Epstein-Barr virus infectionwhereas those in the immunocompetent populationare not.

Diagnosis

PCNSL usually occurs in the fifth and sixth decades oflife. Neurologic symptoms and signs depend on thesite(s) of disease in the brain, but cognitive changesand lateralizing signs are common. PCNSL arises in

the basal ganglia, corpus callosum, and periventricu-lar regions. Following gadolinium administration,these tumors show intense and distinctive homoge-neous enhancement on MR scans. Because glucocor-ticoids alone are sufficient to reverse vascular per-meability of the tumor and lyse tumor cells, they mustbe withheld until tissue has been obtained to make adefinitive diagnosis. An open or stereotactic biopsy isrequired to establish the diagnosis of PCNSL, but re-section has no therapeutic role in this disease.

Initial Management

Radiation therapy alone is palliative with local con-trol rates of 39% and a median survival of less than1 year after 60 Gy (Nelson, 1999). Conventional sys-temic lymphoma drug combinations such as cy-clophosphamide, doxorubicin, vincristine, and pred-nisone (CHOP) are ineffective (Mead et al., 2000).At present, high-dose methotrexate (HD-MTX) is thesingle most active agent for the treatment of PCNSL.HD-MTX combined with cranial irradiation yields amedian survival of 60 months (Abrey, 2000), but hasbeen associated with neurotoxicity in a significantproportion of patients, particularly those over the ageof 60 at the time of treatment. Almost 100% of suchpatients develop severe dementia (Abrey, 1998).

Efforts to overcome CNS toxicity from irradiationhas led to the use of chemotherapy alone, particu-larly in older patients. Using an HD-MTX based regi-men, patients older than 60 years have the same me-dian survival (32 months) as those treated with thesame regimen plus cranial radiotherapy. However,no neurotoxicity was observed in those who only re-ceived chemotherapy. Intrathecal chemotherapy hasnot attained a defined role in PCNSL management asdata indicate that treatment with HD-MTX alone pro-duces comparable results in patients with a negativeCSF cytologic examination at diagnosis. Those with apositive CSF cytology should receive concurrent in-trathecal and systemic chemotherapy. Chemotherapycombined with blood-brain barrier disruption hasbeen another approach; in one study, 74 patients hadan estimated median survival of 40.7 months. Of 36patients with a complete response lasting more than1 year and available for study, none demonstrated ev-idence of cognitive loss in neuropsychologic testsand/or clinical examinations (McAllister et al., 2000).

Most neuro-oncologists agree that the optimaltreatment for PCNSL has not yet been identified. In arecent review of published clinical trials, Ferreri and


3601_e15_p362-374 2/19/02 8:56 AM Page 372

colleagues (Ferreri et al., 2000) found that chemo-therapy followed by radiotherapy yielded a 5-year sur-vival of 22%–40% compared with 3%–26% for irra-diation only. In their review, HD-MTX was the mosteffective chemotherapy, producing response rates of80%–90% and a 2-year survival rate of 60%–65%.To date, the addition of other drugs at conventionaldoses for treatment has not consistently improvedoutcome. With a few exceptions, any regimen with-out HD-MTX performed no better than RT alone.

REFERENCES

Abrey LE, DeAngelis LM, Yahalom J. 1998. Long-term survivalin primary CNS lymphoma. J Clin Oncol 16:859–863.

Abrey LE, Yahalom J, DeAngelis LM. 2000. Treatment for pri-mary CNS lymphoma: the next step. J Clin Oncol 18:3144–3150.

Balm M, Hammack J. 1996. Leptomeningeal carcinomatosis.Presenting features and prognostic factors. Arch Neurol53:626–632.

Barcos M, Lane W, Gomez GA, et al. 1987. An autopsy studyof 1206 acute and chronic leukemias (1958 to 1982).Can-cer 60:827–837.

Berg SL, Balis FM, Zimm S, et al. 1992. Phase I/II trial andpharmacokinetics of intrathecal diaziquone in refractorymeningeal malignancies. J Clin Oncol 10:143–148.

Berweiler U, Krone A, Tonn JC. 1998. Reservoir systems forintraventricular chemotherapy. J Neurooncol 38:141–143.

Bilsky MH, Lis E, Raizer J, Lee H, Boland P. 1999. The diag-nosis and treatment of metastatic spinal tumor. Oncologist4:459–469.

Bleyer WA, Poplack DG. 1979. Intraventricular versus intra-lumbar methotrexate for central-nervous-system leukemia:prolonged remission with the Ommaya reservoir. Med Pe-diatr Oncol 6:207–213.

Bower JH, Hammack JE, McDonnell SK, Tefferi A. 1997. Theneurologic complications of B-cell chronic lymphocytic leu-kemia. Neurology 48:407–412.

Byrne TN, Waxman SG. 1990. Spinal Cord Compression: Di-agnosis and Principles of Management. Philadelphia: FADavis, 278 pp.

Castagnola C, Nozza A, Corso A, Bernasconi C. 1997. The valueof combination therapy in adult acute myeloid leukemiawith central nervous system involvement. Haematologica82:577–580.

Chamberlain MC. 1998. Radioisotope CSF flow studies in lep-tomeningeal metastases. J Neurooncol 38:135–140.

Champagne MA, Silver HK. 1992. Intrathecal dacarbazine treat-ment of leptomeningeal malignant melanoma. J Natl Can-cer Inst 84:1203–1204.

Cibas ES, Malkin MG, Posner JB, Melamed MR. 1987. Detec-tion of DNA abnormalities by flow cytometry in cells fromcerebrospinal fluid. Am J Clin Pathol 88:570–577.

Davis FG, Preston-Martin S. 1999. Epidemiology. Incidenceand survival. In: Bigner DD, McLendon RE, Bruner JM(eds), Russell and Rubinstein’s Pathology of Tumors of theNervous System, London: Arnold, pp. 5–46.

DeAngelis LM. 1998. Current diagnosis and treatment of lep-tomeningeal metastasis. J Neurooncol 38:245–252.

Ferreri AJ, Reni M, and Villa E. 2000. Therapeutic manage-ment of primary central nervous system lymphoma: lessonsfrom prospective trials. Ann Oncol 11:927–937.

Freilich RJ, Krol G, DeAngelis LM. 1995. Neuroimaging andcerebrospinal fluid cytology in the diagnosis of lep-tomeningeal metastasis. Ann Neurol 38:51–57.

Gilbert RW, Kim JH, Posner JB. 1978. Epidural spinal cordcompression from metastatic tumor: diagnosis and treat-ment. Ann Neurol 3:40–51.

Glantz MJ, Cole BF, Recht L, et al. 1998. High-dose intravenousmethotrexate for patients with nonleukemic leptomeningealcancer: is intrathecal chemotherapy necessary? J Clin On-col 16:1561–1567.

Glantz MJ, Hall WA, Cole BF, et al. 1995. Diagnosis, manage-ment, and survival of patients with leptomeningeal cancerbased on cerebrospinal fluid-flow status. Cancer 75:2919–2931.

Glantz MJ, LaFollette S, Jaeckle KA, et al. 1999. Randomizedtrial of a slow-release versus a standard formulation of cy-tarabine for the intrathecal treatment of lymphomatousmeningitis. J Clin Oncol 17:3110–3116.

Gray JR, Wallner KE. 1990. Reversal of cranial nerve dysfunc-tion with radiation therapy in adults with lymphoma andleukemia. Int J Radiat Oncol Biol Phys 19:439–444.

Hanssens PE, Lagerwaard FJ, Levendag PC. 1998. Principles ofradiotherapy of neoplastic meningosis. J Neurooncol38:145–150.

Jordan JE, Donaldson SS, Enzmann DR. 1995. Cost effective-ness and outcome assessment of magnetic resonance imag-ing in diagnosing cord compression. Cancer 75:2579–2586.

Kataoka A, Shimizu K, Matsumoto T, et al. 1995. Epiduralspinal cord compression as an initial symptom in child-hood acute lymphoblastic leukemia: rapid decompressionby local irradiation and systemic chemotherapy. PediatrHematol Oncol 12:179–184.

Klekamp J, Samii H. 1998. Surgical results for spinal metas-tases. Acta Neurochir (Wien) 140:957–967.

Larson SM, Schall GL, DiChiro G. 1971. The influence of pre-vious lumbar puncture and pneumoencephalography on theincidence of unsuccessful radioisotope cisternography. JNucl Med 12:555–557.

Levitt LJ, Dawson DM, Rosenthal DS, Moloney WC. 1980. CNSinvolvement in the non-Hodgkin’s lymphomas. Cancer45:545–552.

Loblaw DA, Laperriere NJ. 1998. Emergency treatment of ma-lignant extradural spinal cord compression: an evidence-based guideline. J Clin Oncol 16:1613–1624.

Lossos IS, Breuer R, Intrator O, Lossos A. 2000. Cerebrospinalfluid lactate dehydrogenase isoenzyme analysis for the di-agnosis of central nervous system involvement in hema-tooncologic patients. Cancer 88:1599–1604.

Lyons MK, O’Neill BP, Marsh WR, Kurtin PJ. 1992. Primaryspinal epidural non-Hodgkin’s lymphoma: report of eightpatients and review of the literature. Neurosurgery30:675–680.

Mackintosh FR, Colby TV, Podolsky WJ, et al. 1982. Centralnervous system involvement in non-Hodgkin’s lymphoma:an analysis of 105 cases. Cancer 49:586–595.


3601_e15_p362-374 2/19/02 8:56 AM Page 373

Mahmoud HH, Rivera GK, Hancock ML, et al. 1993. Low leuko-cyte counts with blast cells in cerebrospinal fluid of chil-dren with newly diagnosed acute lymphoblastic leukemia.N Engl J Med 329:314–319.

Maranzano E, Latini P, Beneventi S, et al. 1996. Radiotherapywithout steroids in selected metastatic spinal cord com-pression patients. A phase II trial. Am J Clin Oncol19:179–183.

Maranzano E, Latini P, Checcaglini F, et al. 1991. Radiationtherapy in metastatic spinal cord compression. Cancer 67:1311–1317.

Mason WP, Yeh SDJ, DeAngelis LM. 1998. 111Indium-diethyl-enetriamine pentaacetic acid cerebrospinal fluid flow stud-ies predict distribution of intrathecally administered che-motherapy and outcome in patients with leptomeningealmetastases. Neurology 50:438–444.

McAllister LD, Doolittle ND, Guastadisegni PE, et al. 2000. Cog-nitive outcomes and long-term follow-up results after en-hanced chemotherapy delivery for primary central nervoussystem lymphoma. Neurosurgery 46:51–60.

Mead GM, Bleehen NM, Gregor A, et al. 2000. A medical re-search council randomized trial in patients with primarycerebral non-Hodgkin lymphoma: cerebral radiotherapywith and without cyclophosphamide, doxorubicin, vin-cristine, and prednisone chemotherapy. Cancer 89:1359–1370.

Moser AM, Adamson PC, Gillespie AJ, Poplack DG, Balis FM.1999. Intraventricular concentration times time (C � T)methotrexate and cytarabine for patients with recurrentmeningeal leukemia and lymphoma. Cancer 85:511–516.

Nelson DF. 1999. Radiotherapy in the treatment of primarycentral nervous system lymphoma (PCNSL). J Neurooncol43:241–247.

Olson ME, Chernik NL, Posner JB. 1974. Infiltration of the lep-tomeninges by systemic cancer. Arch Neurol 30:122–137.

Oschmann P, Kaps M, Volker J, Dorndorf W. 1994. Meningealcarcinomatosis: CSF cytology, immunocytochemistry andbiochemical tumor markers. Acta Neurol Scand 89:395–399.

Oviatt DL, Kirshner HS, Stein RS. 1982. Successful chemother-apeutic treatment of epidural compression of non-Hodg-kin’s lymphoma. Cancer 49:2446–2448.

Pfefferbaum A, Mathalon DH, Sullivan EV, Rawles JM, ZipurskyRB, Lim KO. 1994. A quantitative magnetic resonance imag-ing study of changes in brain morphology from infancy tolate adulthood. Arch Neurol 51:874–887.

Posner JB. 1987. Back pain and epidural spinal cord com-pression. Med Clin North Am 71:185–205.

Posner JB. 1995. Neurologic Complications of Cancer. Phila-delphia: FA Davis, 482 pp.

Posner JB, Howieson J, Cvitkovic E. 1977. “Disappearing”spinal cord compression: oncolytic effect of glucocorticos-teroids (and other chemotherapeutic agents) on epiduralmetastases. Ann Neurol 2:409–413.

Price RA, Johnson WW. 1973. The central nervous system inchildhood leukemia: I. The arachnoid. Cancer 31:520–533.

Raz I, Siegal T, Siegal T, Polliack A. 1984. CNS involvement bynon-Hodgkin’s lymphoma. Response to a standard thera-peutic protocol. Arch Neurol 41:1167–1171.

Recht L, Straus DJ, Cirrincione C, Thaler HT, Posner JB. 1988.Central nervous system metastases from non-Hodgkin’s

lymphoma: treatment and prophylaxis. Am J Med 84:425–435.

Rhodes CH, Glantz MJ, Glantz L, et al. 1996. A comparison ofpolymerase chain reaction examination of cerebrospinalfluid and conventional cytology in the diagnosis of lym-phomatous meningitis. Cancer 77:543–548.

Rodesch G. Van Bogaert P, Mavroudakis N, et al. 1990. Neu-roradiologic findings in leptomeningeal carcinomatosis: thevalue interest of gadolinium-enhanced MRI. Neuroradiol-ogy 32:26–32.

Rogers LR, Duchesneau PM, Nunez C, et al. 1992. Compari-son of cisternal and lumbar CSF examination in lep-tomeningeal metastasis. Neurology 42:1239–1241.

Schiff D, Shaw EG, Cascino TL. 1995. Outcome after spinalreirradiation for malignant epidural spinal cord compres-sion. Ann Neurol 37:583–589.

Schold SC, Wasserstrom WR, Fleisher M, Schwartz MK, Pos-ner JB. 1980. Cerebrospinal fluid biochemical markers ofcentral nervous system metastases. Ann Neurol 8:597–604.

Shapiro WR, Young DF, Mehta BM. 1975. Methotrexate: dis-tribution in cerebrospinal fluid after intravenous, ventricu-lar and lumbar injections. N Engl J Med 293:161–166.

Siegal T. 1998. Leptomeningeal metastases: rationale for sys-temic chemotherapy or what is the role of intra-CSF-che-motherapy? J Neurooncol 38:151–157.

Siegal T, Lossos A, Pfeffer MR. 1994. Leptomeningeal metas-tases: analysis of 31 patients with sustained off-therapy re-sponse following combined-modality therapy. Neurology44:1463–1469.

Sioutos PJ, Arbit E, Meshulam CF, Galicich JH. 1995. Spinalmetastases from solid tumors. Analysis of factors affectingsurvival. Cancer 76:1453–1459.

Stark B, Sharon R, Rechavi G, et al. 2000. Effective preventivecentral nervous system therapy with extended triple in-trathecal therapy and the modified ALL-BFM 86 chemo-therapy program in an enlarged non-high risk group of chil-dren and adolescents with non-B-cell acute lymphoblasticleukemia: the Israel National Study report. Cancer 88:205–216.

Stockhammer G, Poewe W, Burgstaller S, et al. 2000. Vas-cular endothelial growth factor in CSF. A biologicalmarker for carcinomatous meningitis. Neurology 54:1670–1676.

Tubergen DG, Cullen JW, Boyett JM, et al. 1994. Blasts in CSFwith a normal cell count do not justify alteration of therapyfor acute lymphoblastic leukemia in remission: a ChildrensCancer Group Study. J Clin Oncol 12:273–278.

van der Gaast A, Sonneveld P, Mans DR, Splinter TAW. 1992.Intrathecal administration of etoposide in the treatment ofmalignant meningitis: feasibility and pharmacokinetic data.Cancer Chemother Pharmacol 29:335–337.

Van Oostenbrugge RJ, Hopman AH, Ramaekers FC, Twijn-stra A. 1998. In situ hybridization: a possible diagnosticaid in leptomeningeal metastasis. J Neurooncol 38:127–133.

Wasserstrom WR, Glass JP, Posner JB. 1982. Diagnosis andtreatment of leptomeningeal metastases from solid tumors:experience with 90 patients. Cancer 49:759–772.

Wong ET, Portlock CS, O’Brien JP, DeAngelis LM. 1996.Chemosensitive epidural spinal cord disease in non-Hodg-kin’s lymphoma. Neurology 46:1543–1547.


3601_e15_p362-374 2/19/02 8:56 AM Page 374

Documents

Leukemia and Lymphoma Metastases - Society for Neuro-Oncology - a